AI-200: Azure AI Cloud Developer Associate Exam Prep

Explore core generative AI jargons, including tokens, system and user prompts, and the chat completions API, and compare multimodal and unimodal models for robust Azure AI development.

Deploy a Microsoft Foundry resource with a new resource group and project, then use the Foundry portal to access model catalog, endpoints, and AI services.

Explore how to call a large language model deployed in a Microsoft Foundry resource using the Foundry SDK, environment variables, and an OpenAI client to run chart completions.

Create your first Foundry agent, apply versioning and guardrails, attach tools like web search and code interpreter, and connect to enterprise knowledge via FoundryIQ with memory store.

Install and run Docker Desktop to containerize your AI application for labs. Follow the installation guide and log in to build and push a container image to Azure resources.

Configure a container health check in your Azure App Service plan by pinging the /health endpoint, monitoring unhealthy instances, and automatically removing them from the load balancer.

Discover how to use the kudu console to inspect an azure web app's persistent storage, view environment variables and logs, and access persisted json files in the /home/chat_history folder.

Explore deployment slots in Azure web apps to host production, staging, and development environments, swap slots with near zero downtime, and integrate with ci/cd workflows and container registries.

Connect to your Azure AKS cluster, using Azure CLI credentials, and inspect nodes, pods, and namespaces with kubectl. Install kubectl and verify the version to interact with kube-system namespaces.

Learn how Azure Kubernetes Service uses config maps for endpoint and model name and secrets for API key to enable runtime injection and modularity.

Explore persistent volumes and persistent volume claims in AKS, using Azure file share to retain data across pod restarts, with YAML manifests and static and dynamic provisioning.

Learn to use environment variables and secrets to power a Python app in Azure container apps, including dockerization, Azure container registry deployment, and secret references to OpenAI endpoints.

Store the API key in Azure Key Vault and link it to an Azure container app using a user assigned managed identity. The app fetches key via a secret reference.

Azure Container Apps event-driven auto-scaling using HTTP scaler and blob storage rules, scaling replicas from one to ten during spikes and back, powered by KEDA.

Explore how to boost Azure Cosmos DB query performance by implementing range and composite indexes, optimizing filtering and sorting on rating, review count, and category.

Explore vector storage design in Azure Cosmos DB, including embedding field structure, vector embedding policy, dimensions, float32 data type, cosine similarity, and disk ANN indexing.

Integrate vector embeddings with a JSON data set in an Azure Cosmos DB container, using a vector policy and disk ANN index for retrieval augmented generation.

Explore the ESG reporting case study for labs, modeling environmental emissions data in Azure managed Postgres SQL, covering scope one to three, indexing, queries, and rack chatbot with vector embeddings.

Create and manage a PostgreSQL database in Azure by building an ESG schema with tables for companies, emissions, energy use, and sustainability reports, then insert and query sample data.

Demonstrate a code-first connection to Azure PostgreSQL with the PsychoPG SDK, configuring environment variables, creating a connection pool, and running selects with where, order by, and aggregations.

Learn chunking strategies for a RAG pipeline, including fixed size with overlaps, semantic, and context enriched approaches, to preserve meaning and optimize vector embeddings and retrieval.

Enable the vector extension in Azure PostgreSQL to store vector embeddings and set up a RAG schema with ESG text data and ESG chunks tables, including a 1536-dimension vector column.

Deploy an Azure managed Redis instance in the Azure portal with an in-memory data tier, 0.5 gb, 2 vcores, p0sku, public networking, and clustering for vector storage and vector search.

Execute a hands-on lab to connect to an Azure managed Redis instance by configuring hostname and password environment variables, and perform set and get, hash operations, and pipeline executions.

Compare flat and hnsw vector indexing in Redis, noting accuracy and latency tradeoffs; outline embedding storage in a Redis hash with cosine similarity and ef runtime.

Upload a dataset to a redis instance with 1536-d vector embeddings, store them as redis hashes, and create an hnsw vector index to enable semantic search with the Ada002 model.

Run vector search similarity queries on Redis with embeddings and a KNN HNSW index; learn embedding generation, binary conversion, and cosine distance-based ranking.

Explore the Azure Service Bus message structure, including the request payload, message ID, content type, correlation ID, and application properties for filtering and for guiding large language model calls.

Explore Azure service bus message settlement types—complete, abandoned, dead-letter, and defer—and learn how each handles success, transient failures, and manual remediation in a peak lock queue system, including dead-letter queues.

Process dead-lettered messages from a primary queue using a dead-letter queue receiver in a service bus, parsing JSON payloads and invoking an Azure OpenAI model deployment to generate responses.

Set up an Azure Service Bus topic and two subscriptions with SQL filters to route messages to the Microsoft and ESG assistants, and configure topic creation, filters, and message routing.

Implement the claim check pattern with Azure Service Bus to handle large payloads by storing images in Blob storage and sending image URLs in json for processing.